Author + information
- Received January 27, 1997
- Revision received August 7, 1997
- Accepted September 29, 1997
- Published online January 1, 1998.
- William S Weintraub, MD, FACCA,* (, )
- Bernardo Stein, MD, FACCA,
- Andrzej Kosinski, PhDA,
- John S Douglas, Jr., MD, FACCA,
- Ziyad M.B Ghazzal, MD, FACCA,
- Ellis L Jones, MD, FACCA,
- Douglas C Morris, MD, FACCA,
- Robert A Guyton, MD, FACCA,
- Joseph M Craver, MD, FACCA and
- Spencer B King, III, MD, FACCA
- ↵*Dr. William S. Weintraub, Division of Cardiology, Emory University Hospital, 1365 Clifton Road NE, Atlanta, Georgia 30322.
Objectives. This study sought to compare the outcome of percutaneous transluminal coronary angioplasty (PTCA) (n = 834) and coronary artery bypass graft surgery (CABG) (n = 1805) in diabetic patients with multivessel coronary disease from an observational database.
Background. There is concern about selection of revascularization in diabetic patients with multivessel coronary artery disease.
Methods. Data were collected prospectively and entered into a computerized database. Follow-up was by letter or telephone or additional events resulting in readmission.
Results. After CABG there were more in-hospital deaths (0.36% vs. 4.99%, p < 0.0001) and a trend toward more Q wave myocardial infarctions than after PTCA. Five- and 10-year survival rates were 78% and 45% after PTCA and 76% and 48% after CABG, respectively (p = 0.47). At 5 and 10 years, insulin-requiring patients had lower survival rates of 72% and 31% after PTCA and 70% and 48% after CABG, respectively (p = 0.54). Multivariate correlates of long-term mortality were older age, low left ventricular ejection fraction, heart failure and hypertension. In the total group, insulin requirement was a correlate of long-term mortality. For the total group, choice of therapy had a multivariate hazard ratio close to 1. In the insulin-requiring subgroup, the multivariate hazard ratio was 1.35 (95% confidence interval 1.01 to 1.79) for PTCA versus CABG. Corrected for baseline differences, 5- and 10-year survival rates were 68% and 36% after PTCA and 75% and 47% after CABG, respectively, in the insulin-requiring subgroup. Nonfatal events were more common after PTCA, especially additional revascularization.
Conclusions. This study reveals a high incidence of events in diabetic patients and raises further questions about angioplasty in insulin-requiring diabetic patients with multivessel disease.
Patients with diabetes mellitus have a marked propensity for cardiovascular morbidity and mortality. Epidemiologic data from the Framingham Study demonstrated a twofold to threefold increase in atherosclerotic disease in diabetic patients. The risk of death from cardiovascular causes is three times higher for diabetic than nondiabetic men, even after adjustment for age and presence of hypercholesterolemia, hypertension and tobacco use . Similarly, the risk of myocardial infarction, peripheral arterial disease and heart failure, and the mortality associated with coronary ischemic events, are increased significantly in diabetic patients .
Because of the prevalence of extensive atherosclerotic disease, diabetic patients constitute an important segment of the population undergoing coronary revascularization procedures. Specifically, patients with diabetes mellitus account for ∼20% of patients currently undergoing revascularization procedures [4–6]. Revascularization of diabetic patients poses a challenge because long-term event rates are known to be higher than in nondiabetic patients [4, 5, 7]. Furthermore, recent data from the Bypass Angioplasty Revascularization Investigation (BARI) and the Coronary Angioplasty Versus Revascularization Investigation (CABRI) have suggested that diabetic patients treated with oral hypoglycemic agents or insulin have a higher mortality with coronary angioplasty than with bypass surgery. There has been concern about the ability to generalize these data from randomized trials to other patient groups because the patients in the randomized trials may not be entirely representative of broader patient populations. Thus, it is necessary to examine observational databases to place the randomized trial data in the proper context.
From January 1981 through December 1994, 2,639 diabetic patients with two- or three-vessel coronary artery disease and no previous revascularization procedures underwent either coronary angioplasty (n = 834) or bypass surgery (n = 1,805) performed at Emory University Hospitals. Included were patients undergoing procedures for stable or unstable angina pectoris or after several days of stabilization after acute myocardial infarction. Those who had the procedure performed emergently in the setting of infarction were excluded.
1.2 Definition of Diabetes
The definition of diabetes follows that of the American College of Cardiology (ACC) database. Although patients were included in the present study before the creation of the ACC database, the definition for the Emory database has always been consistent with ACC database. Patients were classified as having diabetes on the basis of history, regardless of duration of disease or need for antidiabetic agents. The diagnosis could be based on a previous physician telling a patient that they had diabetes (in principle, based on fasting, nonstressed blood sugar level >140 mg/dl on at least two occasions, although this cannot reasonably be verified) or taking or having previously taken oral hypoglycemic agents or insulin or receiving diet therapy.
1.3 Other Definitions
Single-vessel disease= ≥50% diameter lumen narrowing in either the left anterior descending, left circumflex or right coronary artery or a major branch or branches; double-vessel disease= presence of >50% diameter lumen narrowing in two of the three major epicardial vessel systems; three-vessel disease= presence of >50% diameter lumen narrowing in all three major epicardial vessel systems or in the left anterior descending and proximal circumflex coronary artery in left-dominant patients; left main coronary artery disease= presence of >50% diameter lumen narrowing in the left main coronary artery; urgent procedure= a procedure judged by the operator to be required within 24 h; emergent procedure= procedure performed in the setting of acute ischemia or infarction; complete revascularization= all major obstructions in the epicardial vessels are bypassed after bypass surgery or successfully dilated by at least 20% and to <50% diameter stenosis after angioplasty; myocardial infarction= postprocedural myocardial infarction determined by the development of new Q waves; variables defined by patient history= hypertension, severity of angina, previous myocardial infarction or myocardial infarction during follow-up. Anginawas defined by Canadian Cardiovascular Society Classification and congestive heart failureby New York Heart Association criteria .
1.4 Angioplasty and Bypass Surgery Procedures
All angioplasty procedures were performed using standard techniques in use during the time that the procedure was performed and have previously been described . Angioplasty procedures included balloon dilation and stent implantation and the use of other new devices. Because of the period during which most of these patients were treated, the majority of the patients treated with angioplasty underwent balloon angioplasty. For bypass surgery, standard surgical techniques, extracorporeal circulation and myocardial protection methods were used .
1.5 Data Collection
Baseline and restudy demographic, clinical, angiographic and procedural data, including complications, were recorded prospectively on standardized forms and entered into a computerized data base. All fields are defined in a data dictionary. Angiographic success was assessed by quantitative coronary arteriography performed by experienced angiographers involved in the procedure but other than the primary operator.
1.6 Patient Follow-up
Follow-up information was obtained from the patients or their referring physicians. Follow-up status for each end point was also assessed at each subsequent hospital admission. Patients not readmitted were contacted by telephone or letter. Follow-up data were available for 2,541 (96%) of the 2,639 patients. The mean length of follow-up was 5.0 ± 3.6 years. Information obtained included occurrence of myocardial infarction since the initial angioplasty, subsequent need for an additional revascularization procedure (angioplasty or bypass surgery), death (cardiac plus noncardiac) and recurrent angina. All follow-up information was recorded on standardized forms and entered into the computerized data base. All repeat procedures performed at Emory University Hospitals were confirmed from the database. Myocardial infarctions during follow-up were ascertained largely from the patients, and thus there may be both under and overreporting.
1.7 Statistical Analyses
Results are expressed as proportions or mean value ± SD. Differences in categoric variables were analyzed by chi-square analysis (or Fisher exact test), and differences in continuous variables were analyzed by Student ttests. The clinical, angiographic and procedural characteristics of each group were determined. Overall survival (cardiac plus noncardiac) and event-free survival were determined by the Kaplan-Meier method , and the estimated probability is shown together with the standard error of the estimate. Overall survival and event-free survival analyses were performed for the total population, as well as on the insulin-requiring subgroup. End points analyzed included 1) survival; 2) freedom from myocardial infarction; 3) freedom from additional bypass surgery; 4) freedom from additional angioplasty; 5) event-free survival, defined as freedom from the events of death, myocardial infarction, additional bypass surgery or additional angioplasty. Comparisons of total and event-free survival were made using the Mantel-Cox method . Multivariate correlates of survival were analyzed with Cox model analysis with repeated analyses with variables that were frequently missing eliminated to determine the impact of missing data (and consequently fewer patients in the model) on the analyses.
2.1 Coronary Angioplasty Versus Bypass Surgery
The clinical characteristics of all 834 patients treated with coronary angioplasty are compared with those of the 1,805 patients treated with bypass surgery in Table 1. There was no difference in age, with a mean age in the early 60s in both groups. More of the angioplasty group patients were women. There was no difference in the incidence of hypertension. More of the angioplasty group patients had severe angina, whereas more of the bypass surgery group patients had heart failure and a previous myocardial infarction. There was a large difference in vessels diseased, with the majority of the angioplasty group patients having two-vessel disease and the majority of the bypass surgery group patients having three-vessel disease. Left ventricular ejection fraction was in the low normal range and was slightly higher in the angioplasty group patients. Emergent procedures were unusual but were more common in the bypass surgery group patients. In-hospital events and angina during follow-up are shown in Table 2. Completeness of revascularization was much higher with bypass surgery. Just over 3% of the angioplasty group patients crossed over to bypass surgery during the initial hospital period. There were more deaths and a trend toward more Q wave myocardial infarctions in the bypass surgery group. Recurrent angina was noted in just over one third of each group at follow-up.
2.2 Coronary Angioplasty Versus Bypass Surgery in Insulin-Requiring Patients
The clinical characteristics of the 279 insulin-requiring patients treated with coronary angioplasty are compared with those of the 610 insulin-requiring patients treated with bypass surgery in Table 3. The patients are similar to the total cohort. There was a trend to younger age in the angioplasty group, with the mean age of both groups in the early 60s. There was little difference in gender. There was no difference in the incidence of hypertension. There was a trend toward more of the angioplasty group patients having severe angina, whereas more of the bypass surgery group patients had heart failure. There was a large difference in vessels diseased, with the majority of the angioplasty group patients having two-vessel disease and the majority of the bypass surgery group patients having three-vessel disease. Left ventricular ejection fraction was in the low normal range and was slightly higher in the angioplasty group. Emergent procedures were unusual but were more common in the bypass surgery group patients. In-hospital events and angina during follow-up are shown in Table 4. Completeness of revascularization was much higher with bypass surgery. Just over 5.5% of the angioplasty group patients crossed over to bypass surgery during the initial hospital period. There were more deaths in the surgery group. There was a slight trend toward more angina during follow-up in the angioplasty group.
Survival for all diabetic patients is displayed in Fig. 1. Note that initially the surgical curve drops due to the initial in-hospital mortality. Thereafter, the curves come together and largely overlap. The curves display continuing mortality such that by 10 years, <50% of patients are still alive. Survival in the insulin-requiring subset is similar (Fig. 2), with somewhat a lower survival than in the group overall, as shown in the multivariate analysis that follows. The curves are again seen to largely overlap. Multivariate correlates of long-term mortality are shown in Fig. 3, with the correlates for all patients on top and for the insulin-requiring subgroup below. Older age, low left ventricular ejection fraction, heart failure and hypertension were correlates in the total population and in the insulin-requiring subgroup, although with somewhat wider confidence intervals in the latter. In the total group, insulin requirement was a correlate of long-term mortality as well. For the total group, the choice of therapy had a hazard ratio close to 1, with the 95% confidence interval widely spanning 1. This means that there is no evidence of a survival advantage of one form of therapy over the other. For the insulin-requiring subgroup, the results for the choice of therapy are less clear. Although the survival curves overlapped, the bypass surgery group patients were sicker. Thus, on multivariate analysis there was higher mortality in the insulin-requiring subgroup for coronary angioplasty than for bypass surgery (hazard ratio [HR] 1.35, 95% confidence interval 1.01 to 1.79, p = 0.045). After correction for baseline differences, the 5- and 10-year survival rates were 68% and 36% after coronary angioplasty and 75% and 47% after bypass surgery, respectively, in the insulin-requiring subgroup (Fig. 4). Of note, the number of vessels diseased was not a multivariate correlate of mortality in either the total group or the insulin-requiring subgroup. The survival rate in the subgroup with two-vessel disease at 5 and 10 years was 76% and 51% for angioplasty and 78% and 49% for bypass surgery, respectively (p = 0.79).
2.4 Freedom From Myocardial Infarction
Freedom from myocardial infarction for all patients is shown in Fig. 5and was somewhat lower in the angioplasty group patients than in the bypass surgery group patients. The results in the insulin-requiring subset were essentially identical (data not shown). The multivariate correlates of myocardial infarction were coronary angioplasty as the choice of therapy (HR 1.85, p < 0.0001) and younger age (HR 1.37/decade, p < 0.001).
2.5 Freedom From Additional Bypass Surgery
Freedom from additional bypass surgery for all patients is shown in Fig. 6. There is considerably lower freedom from additional bypass surgery in the angioplasty group. The highest incidence of bypass surgery in the angioplasty group is in the initial 6 months to 1 year, during the period of maximal restenosis. In contrast, in the bypass surgery group, the incidence is initially low but then increases over time, consistent with graft failure or progression of disease. The results in the insulin-requiring subgroup were similar (data not shown). The results were also similar in the subgroup with two-vessel disease, with 5- and 10-year freedom from bypass surgery of 75% and 50% in the angioplasty group and 97% and 82% in the bypass surgery group, respectively. The multivariate correlates of additional bypass surgery were angioplasty as the choice of therapy (HR 8.4, p < 0.0001) and younger age (HR 1.36/decade, p < 0.0001).
2.6 Incidence of Additional Angioplasty
The incidence of additional angioplasty for all patients is shown in Fig. 7. There was a considerably higher incidence in the angioplasty group, with the highest rate in the first 6 months. The incidence in the bypass surgery group gradually increased over time. The results in the insulin-requiring subgroup were similar (data not shown). The results were also similar in the subgroup with two-vessel disease, with 5- and 10-year freedom from coronary angioplasty of 58% and 46% in the angioplasty group and 93% and 79% in the bypass surgery group, respectively. The multivariate correlates of additional coronary angioplasty were angioplasty as the choice of therapy (HR 7.4, p < 0.0001), higher left ventricular ejection fraction (HR 1.13/10% increase, p = 0.0024) and more severe angina at baseline (HR 1.13/one-point increase in angina class, p = 0.009).
2.7 Incidence of Additional Coronary Angioplasty or Bypass Surgery
The incidence of additional angioplasty or bypass surgery for all patients is shown in Fig. 8. The incidence is considerably higher in the angioplasty group patients and higher in both groups than for either angioplasty or bypass surgery as end points alone. Few surviving patients in the angioplasty group will be free of additional revascularization by 7 or 8 years. The results in the insulin-requiring subgroup were similar (data not shown). Freedom from the events of death, myocardial infarction, additional angioplasty or bypass surgery is shown in Fig. 9. Initially, there was a much higher incidence in the angioplasty group, and almost all angioplasty group patients ultimately had a fatal or nonfatal event. However, the bypass surgery group patients continued to have a high incidence of events over the full 10-year period. The results in the insulin-requiring subgroup were similar (data not shown).
3.1 Review of Results
Perhaps the most important finding of the present studywas that the long-term incidence of both fatal and nonfatal events was high, emphasizing the palliative nature of coronary revascularization. This finding limits the importance of differences between the procedures. Although angioplasty group patients were not as sick as the bypass surgery group patients, and complete revascularization was achieved much more often with bypass surgery, there was no difference in survival, either for the group as a whole or for the insulin-requiring subset. An interesting finding was that after accounting for baseline differences, there was still no difference between the groups overall, but there was higher mortality in the angioplasty group patients in the insulin-requiring subset. Data on nonfatal events revealed a high incidence of additional revascularization procedures, especially in the angioplasty group patients. Because there may be more uncertainty concerning therapy in patients with two-vessel disease, this group was considered separately. Patients with two-vessel disease had a fatal and nonfatal event rate similar to the total population for both types of therapy.
3.2 Previous Descriptive Series
Diabetes has been shown to be a risk factor for increased long-term mortality events in multiple studies, both after coronary angioplasty [4, 5, 17, 18]and bypass surgery [7, 19, 20]. In the study by Stein et al. the 5-year survival rate was 89% in diabetic versus 93% in nondiabetic patients, and freedom from infarction was 81% in diabetic versus 89% in nondiabetic patients. Bypass surgery and repeat coronary angioplasty were required more often in diabetic patients. Survival free of infarction or repeat revascularization was only 36% in diabetic versus 53% in nondiabetic patients, with a marked attrition in the first year after angioplasty, which corresponds to the period when restenosis is most common. In the study by Kip et al. , the 9-year survival rate was 64% in diabetic and 82% in nondiabetic patients. Freedom from infarction or repeat procedures was also lower in diabetic patients. In addition, diabetes has been shown to be a risk factor for restenosis [21–28]. In a previous study of 10,291 patients without and 2,372 with diabetes undergoing bypass surgery, the in-hospital mortality rate was 1.8% without diabetes and 4.2% with diabetes. In that study, diabetic patients were shown to have a lower 5- (74% vs. 87%) and 10-year (50% vs 72%) survival rate than nondiabetic patients.
3.3 Randomized Trial Data
Although the Emory Angioplasty Versus Surgery (EAST) study showed no survival difference between bypass surgery and coronary angioplasty in a small number of treated diabetic patients, recent evidence from the BARI and CABRI studies have suggested that diabetic patients treated with insulin or oral agents are at increased risk compared with all other patients and that the risk is higher in the treated diabetic patients in the angioplasty arm. Thus, in the BARI study at 5 years, the survival rate was 65.5% in the angioplasty group and 80.6% in the bypass surgery group, which compares with 71.6% in the angioplasty group and 70.1% in the bypass surgery group in the present study, or 68% and 75% after accounting for baseline differences. Thus, although the results are not as divergent between treatment arms in the present study as in the BARI study, they are similar in direction and in overall mortality.
3.4 Limitations and Conclusions
The major limitation of data concerning choice of therapy from an observational database is that selection bias cannot be overcome, because the variables that account for selection of therapy may not be included or may not be adequately accounted for. However, randomized data can be limited in generalizability. Thus, observations in a randomized trial may only be true for patients within the trial. Thus, from a medical decision-making point of view it is optimum if the randomized data and observational data are similar. The results of randomized trial data as well as the present study raise concerns about performing angioplasty in treated, and especially in insulin-requiring, diabetic patients with multivessel coronary artery disease. However, a definitive answer that only bypass surgery should be performed in such patients would probably overstate the issue. This is because of concern about generalizing the BARI results to broader populations, selection bias in nonrandomized studies and the wide 95% confidence interval surrounding the odds ratio for angioplasty versus bypass surgery (Fig. 3). However, the data are directionally similar in the BARI study and the present study. Furthermore, the high incidence of nonfatal events in the angioplasty group is also disturbing.
It is worth pointing out that decision making has been away from coronary angioplasty and toward bypass surgery in the majority of these patients in any case. The 279 patients with insulin-requiring diabetes and multivessel disease who underwent angioplasty represent 1.5% of 18,591 patients undergoing angioplasty at Emory during this period. Furthermore, in the insulin-requiring diabetic patients, 31% underwent angioplasty as opposed to 35% of the noninsulin-requiring diabetic patients (p = 0.048). Finally, of patients with insulin-requiring diabetes and three-vessel disease, only 73 (15%) of 411 underwent angioplasty, 5 of whom were enrolled in the EAST trial . Thus, medical decision making has already directed most of the most seriously ill patients with insulin-requiring diabetes away from angioplasty toward bypass surgery. Although the data presented in the present study and results from the randomized trials may not be sufficient to make a definitive recommendation about the choice of revascularization in all treated diabetic patients with multivessel coronary disease, the place of angioplasty in these high risk patients is uncertain.
☆ To discuss this article on-line, visit the ACC Home Page at www.acc.org/membersand click on the JACC Forum
- American College of Cardiology
- Bypass Angioplasty Revascularization Investigation
- Coronary Angioplasty Versus Bypass Revascularization Investigation
- Emory Angioplasty Versus Surgery Trial
- hazard ratio
- Received January 27, 1997.
- Revision received August 7, 1997.
- Accepted September 29, 1997.
- The American College of Cardiology
- Stamler J,
- Vaccaro O,
- Neaton JD,
- Wentworth D
- Uusitupa MIJ,
- Niskanen LK,
- Siitonen O,
- Voutilainen E,
- Pyorala K
- Stein B,
- Weintraub WS,
- Gebhart S,
- et al.
- ↵Kip KE, Faxon DP, Detre KM, et al. for the Investigators of the NHLBI PTCA Registry. Coronary angioplasty in diabetic patients: the National Heart, Lung, and Blood Institute Percutaneous Transluminal Coronary Angioplasty Registry. Circulation 1996;94:1818–25.
- Weintraub WS,
- Stein B,
- Gebhart SP,
- Craver JM,
- Jones EL,
- Guyton RA
- Campeau L
- The Criteria Committee of the New York Heart Association
- Douglas JS JR.,
- King SB III.,
- Roubin GS
- Mantel N,
- Haenszel W
- ↵Cox DR. Regression models and life tables. J R Stat Soc B 1972;34:187–2.
- Mick MJ,
- Piedmonte MR,
- Arnold AM,
- Simpfendorfer C
- Alderman EL, Corley SD, Fisher LD, et al., for the CASS Participating Investigators and Staff. Five-year angiographic follow-up of factors associated with progression of coronary artery disease in the Coronary Artery Surgery Study (CASS). J Am Coll Cardiol 1993;22:1141–54.
- Morris JJ,
- Smith LR,
- Jones RH,
- et al.
- Vandormael MG,
- Deligonul U,
- Kern MJ,
- et al.
- Hollman J,
- Badhwar K,
- Beck GJ,
- Franco I,
- Simpfendorfer C
- Lambert M,
- Bonan R,
- Cote G,
- et al.
- Weintraub WS,
- Kosinski AS,
- Brown CL III.,
- King SB III.